The Color Of Gray-scale Image Of The Algorithm And Implementation,

Image, an essential media in information transfer and exchange, is defined as a substance which is obtained by various observing systems in observing the objective world, and which can cause sight by acting on human eyes either directly or indirectly. Color and gray scale are key factors in determining the presentation of an image, and the former plays an important role in human visual system. To better the visual effect by turning grey images into colored images, the colorization processing of image is to be used.The traditional method of turning a grey image into a colored one is to use pseudo-color processing, which means turning grey image into colored image automatically by establishing a one-to-one relationship between grey scale and color. This method is quick and simple without changing the information of image, but its colorization effect is awkward and unnatural with limited field of application.Afterwards, Ruderman came up with a lαβ color space which is relativity-resistant. That is to say, the three channels in the space are orthometric, which makes it possible to change that color value of each channel. Based on such relativity-resistant color space, Reinhard put forward a color transfer arithmetic, using which we could change the color of an image in order to maintain its original shape with the color of the reference image. Then, Welsh improved this arithmetic and made it applicable to grey image, which means that we could colorize a grey image by using a reference color image.This thesis starts with the introduction of relative knowledge on image processing and then gives a detailed analysis on Reinhard's color transfer arithmetic and Welsh's grey image colorization arithmetic; based on the above theories, the author of this these has come up with a grey image colorization arithmetic based on the pels space relativity, and improved this arithmetic by using blocking processing method. Finally, the author of this thesis proposes a multi-resolution colorization arithmetic in the field of human face image processing which gets fine effect.